Tony

"Texas alone has a larger GDP than Brazil. Florida is about to surpass Brazil too." That's Cristina Junqueira explaining why $NU just filed for a US federal bank charter. The market size is obvious. But here's what makes this interesting: Nubank already has 4-5 million customers from Brazil, Mexico, and Colombia downloading the app and making regular purchases in the US. These will be their early adopters. Starting with the Latino community makes business sense. The US Hispanic population is 68 million people with $4.1 trillion in GDP. About 9.5% are unbanked, another 20% underbanked. That's 6.2 million households underserved by traditional banks. If Nubank captures just 10% in year one, that's 620,000 accounts and roughly $4.8 billion in deposits. Almost what Nu Mexico built in 5 years. But Cristina is clear this isn't just a play for Latinos. She thinks Nubank's sub-$1 cost to serve can work for any American consumer. No fees, transparent pricing, simple mobile experience. Universal pain points that their tech platform has already solved across three markets. Nubank already has 150+ employees in the US from acquisitions and local hiring. Cristina relocated to Miami and will run the US bank as CEO. The regulatory process will take years, but they're getting ahead of it now. Can a neobank that cracked LatAm actually win in the US market?

@myrrakle @RebillTeam

Que paywall que no sea stripe, recomiendan para un saas?

@benln Would love to be included! I want some accountability for the app I am building now

Rebooting my X group chat for people building projects on Sundays. One rule: you can only post on Sundays. Reply if you want in.

@zarazhangrui She is the goat! Her episode on Lenny’s podcast about marketplaces is a masterclass.

How can someone be THIS early

@NickADobos This actually makes sense. A friend created the own agents inside Slack to make code changes based on the team context and feedback. I am sure they saw people doing this at scale, so they implemented this feature.

Anthropic isn't using claude code???

@codyschneider mappa.ai

who's building this "ai recruiting software"

@kunchenguid How is this different from /pr-review? Is it more comprehensive?

/no-mistakes is here! by popular demand i've made the most impactful tool in my agentic engineering setup "no-mistakes" invocable as a skill in Claude Code, Codex et al just type "/no-mistakes" once your agent has made changes, and watch the magic unfold details below 👇

@aakashgupta People forgot that Jason Blum, the maker of Paranormal Activity, is the executive producer of the movie and has actively championed the film's theatrical success. He saw the movie and wanted to be part of it. Legend!

The $750k everyone is dividing by isn't the number that actually made the money. Curry Barker shot Obsession for $750k. Focus then paid around $14M to acquire it at TIFF, and a 2,600-theater wide release runs tens of millions more in prints and advertising. The real cost basis behind that $148M is closer to $50M than $750k. Still one of the best trades in modern Hollywood. Just not the trade the post describes. Two different companies are winning here. Capstone financed the $750k production and sold worldwide rights, so their return is the $14M check plus international. Focus is playing the distribution margin, where studios keep roughly half of box office, so a $300M finish nets them somewhere around $150M against that ~$50M spend. And "most profitable of all time" already has a holder. Paranormal Activity turned $15,000 into $193M, a 12,900x return on production budget. Obsession at $330M on $750k is 440x. Astonishing, and still nearly 30x short of the record. The number that goes viral is the one that ignores who actually paid for it.

@_catwu Love the feature, but it will be great to choose the number of agents deployed to avoid burning unnecessary amounts of tokens per workflow.

Excited to share our most powerful new Claude Code feature: dynamic workflows! Mention "workflow" in a prompt and Claude will dynamically create an orchestration plan that it strictly follows, allowing you to confidently trust that every stage happens in the right order even across 100s of agents.

@weswinder It does not apply to CC. Why are you saying the opposite?

anthropic just downgraded claude code subs programmatic usage previously used your full rate limits which was something like a 5-10X discount vs api costs now ALL programmatic usage is capped at $20/$100/$200 in api costs (a single opus session can easily burn $20) not good

Hace un año descubrí Flobi y la utilizo para reservar partidos de pádel en Buenos Aires. Muy muy buena. Hoy reservé partidos para el finde. Lo nuevo? El pago fue con @RebillTeam

@duborges @sweepbaseHQ Did you try @raincards ? Not an investor, just a fan. They may require some minimum like Pomelo, but it’s worth a try.

@sweepbaseHQ thx man... we are way too small for Pomelo, they also dont settle in stablecoins, which would require us a huge capital in fiat for reserve balance, so it's a no-go for now we are talking with stripe treasury, but probably not what we need yet

two months bootstrapping a neobank fintech: 1) i think our main lesson last month was: don’t rely on a single partner for money orchestration on sales call, they have everything you need, but when time comes, sooo many issues present - some partners have great fees, but very poor api/docs (basically non-implementable, gotta do everything manually) - some have great API, but extremely rigorous compliance (you integrate the whole thing, to only then find out they wont approve your clients) - some have great support, but wont cover all rails your clients need (pix only, no ACH, no swift, no named accounts, no microdeposits, etc) - some have everything you need, but charge a high monthly minimum with 2x fees you got from the other provider - not to mention the perfect partners, but with abusive contracts that makes our lawyer Claude go crazy we spoke with ~15 orchestration partners for USD, BRL, EUR, AED, CHF and CNY, and realized there’s no “one size fits all” unfortunately so we are basically analyzing what each client needs and activating partner XYZ for them - it’s working fine, besides the tech complexity (though i have already fully integrated 8 partners and it's now taking me 1-2 days to get the whole thing up and running, while larger teams would take weeks - that's the power of repetition) 2) we initially decided on using Base USDC for our backend (even though our clients don’t really see anything stablecoin-related on the dashboard, as a product strategy) but some banking partners wouldnt work on that chain, and Solana seemed to have better coverage and speed so we had to build our own bridging process using circle’s CCTP (to make sure no % fees incurred, eating the low fees we charge) - this was pretty cool but very hard to make it work so now we became not only multi-partners but also multi-chains (no USDT support though, that additional 0.1% was a no-go for us) 3) one of our clients (who turns out to be an old friend), brought his while whole Payroll to Dolafy i built a special dashboard area so they could add all employees/collaborators, enter their banking details, salary and pay date then in a single click every month, they would pay that employee anywhere in the world (via ACH, WIRE, PIX or SEPA), using their Dolafy balance had to work for 20 hours the day it went live, but worked really well and their finance team loved it (much more organized and simple than the bulk-payment via CSV they were doing at Husky) the Payroll product itself is not so profitable in comparison to other clients moving millions in one txn, but it was really satisfying to be able to see all these people getting paid by Dolafy 4) the Credit Cards implementation has been my biggest frustration - i even had nightmares with it and woke up many days in terrible mood even though i implemented their whole API in 2 days, it’s been over a month already and their compliance and legal teams keep slowing us down things like: “hey, you cant mention the term unlimited cards, must be multiple cards” and then i fix it in 5 minutes, and they submit for approval that takes 7 days again and then they come back with “you actually cant talk about cards at all on the homepage, must create a separate page for that” “unlimited back and forth” seems a reasonable term to describe them though in the end, the partner changed the whole copy of our site and we still couldnt get the approval after 40 days ughh tbh, im not sure they are that slow because of bad management or because they try to push the onboarding longer to monetize the setup fees (even though they are losing revenue during this period) anyway, i hope we have cards live in the coming days which will be huge to get more clients and make the current ones move more money with Dolafy (unfortunately there are not many BIN Sponsors for Corporate Cards that we could use besides them) 5) my partner is thinking about moving with his wife to Floripa so we can set up a Dolafy office here the magic island is now the center of crypto and fintech companies in Brazil, plus many of our clients are living here (quality of life and safety for the rich) them startup offices are mainly in the SC/Jurere area (which is opposite side from where I live on the island) but it’s cool to be a short drive away to people with high synergy instead of a long flight so yeah, that’s it for now. talk to you again on our 3rd month building a fintech, hopefully after processing millions in Meta / Google Ads txns (that's what our clients need cards for) thanks for following

Stablecoins connecting local RTP networks has become one the fastest growing payment markets in the world. Probably only beat by the growth in Stablecoin backed card settlement.

@wesbos You can do these videos with AI now. I am sure they are using a skill plus some assets.

WHO IS MAING THESE VIDEOS they are so well done. I need to see the process they use to slice up and animate the UI

@jesse_vermeulen Run other Claude session with a different project scope.

honest question: what do people do during the 5-10 min while Claude is running?

@mattiapomelli Gemini looks cleaner, but graph-wise I prefer Opus 4.7. It makes more sense.

Opus 4.7 vs Gemini 3.1 Pro vs GPT 5.4 Same prompt. Which one do you prefer?

@ankurnagpal @haleymbryant Congrats Ankur! Reading your monthly investor updates was inspirational. I love the radical transparency in each email. I actually recommend my other portfolio companies to follow your communications structure. They yield great results so far.

Carry has been acquired by Angellist and Lettuce We started this company 3.5 years ago to help business owners make better financial decisions These two transactions allow us to continue to do this important work in a bigger way Thank you to everyone who supported us ❤️

@MovieEndorser Tough decision, but Project Hail Mary wins by a tiny margin. I think Interstellar was too long. PHM was perfect.

Interstellar or Project Hail Mary?

@trq212 I am using CC every day, so I’d like to participate

I want to do some streams where I work with non-technical people using Claude Code to figure out how they might be able to improve their process. My feeling is that just a few tips could make a big difference in efficiency. Any mutuals interested?

@DiegoARRG DM y me encantaría intercambiar ideas sobre agents!

si alguien quiere link al dashboard con una mini demo a partir de un ecommerce de prueba (datos sintéticos que dirían los pros) que comente DM y se lo envío por privado

obvio este chaval es mucho más inteligente y su empresa un bicho pero nosotros le estamos dando bastante caña a ser una empresa nativa-ia para que la gente tenga más tiempo para pensar aquí un resumen con ejemplos de como lo tenemos montado: 8 agentes + claudes como nodos